Discriminative Analysis of Nonlinear Functional Connectivity in Schizophrenia

Abstract

Dysfunctional integration of distributed brain networks is believed to be the cause of schizophrenia, and resting-state functional connectivity analyses of schizophrenia have attracted considerable attention in recent years. Unfortunately, existing functional connectivity analyses of schizophrenia using fMRI data have been mostly limited to linear associations, while nonlinear relationships are generally overlooked. In this chapter, a novel measure utilizing the extended maximal information coefficient (eMIC) was introduced to construct whole-brain nonlinear functional connections. In conjunction with multivariate pattern classification, the eMIC-based functional connectivity successfully discriminated the schizophrenic patients from healthy controls with relative higher accuracy rate than the linear measure. Based on the classification accuracy, we found that the strength of the identified linear functional connections involved in the classification decreased in patients with schizophrenia, while the strength of identified functional connections determined by the nonlinear measure increased. Further functional network analysis revealed that the changes of the linear and nonlinear connectivity have similar but not completely the same spatial distribution. In short, the classification results suggest that the nonlinear functional connectivity provided useful discriminative power in the diagnostic classification of schizophrenia, and the inverse changes with similar spatial distribution between the linear and nonlinear measure may indicate the underlying compensatory mechanism between the linear and nonlinear functional connectivity and the complex neuronal synchronization in schizophrenia.